Download International Journal of Computer Science and Intelligent

Pooja Dang et al, International Journal of Computer Science and Mobile Computing, Vol.5 Issue.2, February- 2016, pg. 174-179
Available Online at www.ijcsmc.com
International Journal of Computer Science and Mobile Computing
A Monthly Journal of Computer Science and Information Technology
ISSN 2320–088X
IJCSMC, Vol. 5, Issue. 2, February 2016, pg.174 – 179
Detection of Brain Tumor from MRI
Images Using MATLAB
Pooja Dang
M.Tech(CSE) Department of Computer Science
THE NORTHCAP UNIVERSITY
Gurgaon, Haryana
[email protected]
Jyotika Pruthi
Assistant Professor, Department of Computer Science
THE NORTHCAP UNIVERSITY
Gurgaon, Haryana
[email protected]
Abstract— Medical image processing is the most emergent field these days and processing of images scanned
through MRI are the part of it. For image processing, Segmentation is widely used. In that way MRIMagnetic Resonance Imaging has become practical means for the analysis of brain tumor images. In the
paper we have presented some noise removal functions, three types of segmentation and morphological
operations. Using MATLAB software, we have detected and extracted the tumor from MRI scan images.
Keywords- MRI, histogram, segmentation, brain tumor.
I.
INTRODUCTION
Brain cancer is the virus of the brain. Cancer arises basically when the body cells start to divide without any
control and they spread around the nearby parts. Some of the cancers form solid tumors. Tumors also named as
neoplasm are the masses of the tissue. Many medical imaging modalities are available today to capture the
abnormality such as tumor in human body. Magnetic Resonance Imaging (MRI), Computed Tomography (CT)
scan, X-ray, Ultrasound and Electrocardiogram (ECG) are some of the modalities which are used to capture the
images. MRI is best appropriate and believed to be powerful technology among all others to collect ideal internal
data of the human body for clinical diagnosing. In MRI images, the quantity of data is in large amount, that why
the manual interpretation and its analysis becomes difficult and time consuming. From the past few years, brain
tumor segmentation in MRI has turn into a growing study area. For the identification of abnormal tissue (tumor),
location and size of brain tumor plays an important role.
© 2016, IJCSMC All Rights Reserved
174
Pooja Dang et al, International Journal of Computer Science and Mobile Computing, Vol.5 Issue.2, February- 2016, pg. 174-179
We propose an algorithmic approach for tumor (abnormal tissue) recognition based on segmentation and
morphological operators. First of all, the image is enhanced and then various segmentation techniques are applied
on it to detect the tumor part. In future the work will be extended to categorize the particular tissue as either
Normal or Abnormal Tissue using a classification technique called as Support Vector Machine. Then volume of
the extracted tumor region will be calculated to analyze its size.
II.
RESEARCH METHODOLOGY
Using various image processing modalities, we have developed an algorithm for the detection of abnormal mass
of tissue in the brain scanned through MRI. In this, we are presenting a methodology that detects the tumor region
present in the brain.
According to this algorithm we have two stages, namely Preprocessing and Post processing.
The architecture of Methodology is as follows:
Figure 1. Flow Chart of Methodology
A.
Image Acquisition
MRI images are acquired when a patient is scanned by MRI Machine on the computer. MRI images can be
taken of any part of the body which is desired or to be tested. Images are taken on a 1.5T using T1, T2- weighted
axial contrast images. Thickness of the image is slice 0.5 cm, and image format is PNG and JPEG.
B. Preprocessing
Using MATLAB, brain image scanned through MRI is imported using the command “imread()”and displayed
using “imshow()”. Then this image is converted into gray scale image .After this, skull stripping is performed,
that is removal of non-brain regions. This is an important step in preprocessing which is performed using the
morphological operator ” imopen()”.Then we have applied contrast stretching techniques known as Histogram
equalization that stretch out the intensity values. In this step, we are basically concerned with the location of
abnormal tissue present which is found in the bottom region of the scanned image.
© 2016, IJCSMC All Rights Reserved
175
Pooja Dang et al, International Journal of Computer Science and Mobile Computing, Vol.5 Issue.2, February- 2016, pg. 174-179
Original Image
Grayscale image
Skull Stripped Image
Contrast Enhanced Image
Cropped Image
Figure 2. Preprocessing steps
C. Postprocessing
1. Image division
In this module, the image is divided into two halves as top and bottom to extract features from each half.
The following algorithm is used to accomplish this procedure:
i)Take two variables as A and B for storing the number of rows and columns.
ii) Take some other variable C that contains the number of columns divided by 2.
iii) Using the command “imcrop()”,image will be the half of the input image.
2. Histogram Plotting and comparison between them
In digital imaging world, Histogram Plotting is a graphical depiction between scale of pixels and the intensity of
pixels of the image. In the figure, the horizontal axis shows the scale of pixel intensity and vertical axis shows the
intensity of pixel present in the image. Figure 3 (a), (b), (c) shows the histogram graph of the original image and
the plot of both halves. It is clearly visible that the bottom half has more deviation around 200.Figure 4 indicates
that abnormal tissue is present in the bottom part of the image.
4000
3500
3000
2500
2000
1500
1000
500
0
0
50
100
150
200
150
200
250
Original Image and its histogram
x 10
4
7
6
5
4
3
2
1
0
0
50
100
250
Top half and its histogram
© 2016, IJCSMC All Rights Reserved
176
Pooja Dang et al, International Journal of Computer Science and Mobile Computing, Vol.5 Issue.2, February- 2016, pg. 174-179
16000
14000
12000
10000
8000
6000
4000
2000
0
0
50
100
150
200
250
Bottom half and its histogram
Figure 3. Difference of two histograms
3. Compute Threshold Segmentation
Threshold segmentation is the easiest segmentation which is widely used. This method is based on getting
binary image out of gray scale image. To segment abnormal tissue we are using thresholding as segmentation
algorithm. In this first of all, select a threshold value for initial segmentation. Various methods are present for
this task such as K-means clustering, Otsu’s method. Using K-means clustering, representation of an image can
be made which is somewhat easier to analyse the results. This is basically used to locate boundaries and objects.
image labeled by cluster index
Figure 4.
(a) Object in cluster index
(b)objects in cluster
Gradient magnitude (gradmag)
50
100
150
200
250
50
(c)Gradient magnitude
100
150
200
250
(d) After Thresholding Technique
4. Compute Watershed Segmentation
In this transformation, image is considered as a topographic surface and the gray level of the image represents the
altitudes. Regions with a constant gray level constitute the flat area of an image and region edges depict high
watersheds and low rise region depicts catchment basis. Every part has their own minima, and even if single drop
of water falls on one plane of the watershed, minima of that region will be reached. Watershed lines represent the
boundaries of the object.
© 2016, IJCSMC All Rights Reserved
177
Pooja Dang et al, International Journal of Computer Science and Mobile Computing, Vol.5 Issue.2, February- 2016, pg. 174-179
Watershed transform of gradient magnitude (Lrgb)
Figure 5.Watershed tarsnform of gradient magnitude
5. Morphological Operators
After applying threshold and watershed segmentation, chain of morphological operations are applied. In this,
first we have applied erosion operator to shrink the objects in the image. Size and shape of the structuring element
is been controlled by this operation. Then we have applied Dilation operator which will grow or thickens the
object. Structuring element plays an important role in both of the operations. Then we apply the operations such
as opening and closing formed from combining these two (dilation and erosion) operations. After applying all
these operations, we can clearly figure out the tumor (abnormal) region present in the brain.
Figure 5. Tumor Region as final output
III.
APPLICATIONS
The Magnetic Resonance Imaging (MRI) technique is the finest due to its superior resolution than the other
methods. For the early detection of brain tumor images, MRI is widely used. To make complete images of the
body parts and cells within the body, Magnetic Resonance Imaging (MRI) Scan utilizes strong radio waves and
magnetic field. Manual segmentation of the abnormal tissue from the scanned images using MRI is generally
tricky and takes lot of time. That’s why we are using Automated segmentation technique that give suitable results
with less work for the manual operator. The goal of this work is to offer an automated tool which locates the
tumor on MRI scanned image.
IV.
CONCLUSION
The research was done for detection and extraction of tumor from the MRI image. Various noise removal
techniques were used. Mainly segmentation was used to extract the tumor region based on the methods as
Watershed, Texture and Threshold Segmentation. In this study, various samples were collected, scanned using
MRI, preprocessed and further evaluated using morphological operators, thus locating the tumor and getting
efficient end results. In future the work will be extended to generate a new algorithm for volume calculation of
the tumor region to analyze its size.
ACKNOWLEDGEMENT
I would like to express my sincere gratitude to my project guide Ms. Jyotika Pruthi, Assistant Professor,
THE NORTHCAP University, Gurgaon for her guidance and help which helped out to carry this research work.
© 2016, IJCSMC All Rights Reserved
178
Pooja Dang et al, International Journal of Computer Science and Mobile Computing, Vol.5 Issue.2, February- 2016, pg. 174-179
REFERENCES
[1]
[2]
[3]
[4]
[5]
T. Logeswari and M.Karnan “An improved implementation of brain tumor detection using segmentation based on soft computing”
Journal of Cancer Research and Experimental Oncology Vol. 2(1) pp. 006-014, March, 2010
Magnetic Resonance Image, [Online Available], www.CEwebsource.com, accessed on 18 June 2013.
Dibyendu Goshal, Pinaki Pratim Acharjya, “MRI Image Segmentation using Watershed Transform”, International Journal of Emerging
Technology and Advanced Engineering, ISSN 2250-2459, Volume 2, Issue 4, April 2012.
Gopal,N.N. Karnan, M. , ―Diagnose brain tumor through MRI using image processing clustering algorithms such as Fuzzy C Means
along with intelligent optimization techniques ― , Page(s): 1– 4, Computational Intelligence and Computing Research (ICCIC),2010
IEEE International Conference, 28-29 Dec. 2010.
Manor K Kowari and Sourabh Yadav, “Brain Tumor Detection and Segmentation using Histogram Thresholding”, International
Journal of Engineering and Advanced Technology (IJEAT) ISSN: 2249-898, Volume-1, Issue-4, Journal, India, April 2012.
© 2016, IJCSMC All Rights Reserved
179